Inception

Diffusion-based LLMs delivering 5-7x faster inference with frontier quality at lower cost.

Overview

Inception is an AI research and product company building diffusion-based large language models (dLLMs) for production applications. Unlike traditional autoregressive LLMs that generate tokens sequentially, Mercury diffusion models generate tokens in parallel through a coarse-to-fine refinement process, enabling sub-300ms time-to-first-token and 5-7x higher throughput at up to 70% lower cost.

The company's approach treats AI generation more like an editor than a one-way typewriter, allowing fine-grained control over outputs, adherence to specific schemas, and a unified paradigm for combining language with other data modalities like audio, images, and video. Mercury models are OpenAI API compatible and available through major cloud providers including AWS Bedrock and Azure Foundry.

Key features

Parallel token generation
Sub-300ms time-to-first-token
5-7x higher throughput than autoregressive models
OpenAI API compatible
Schema and constraint adherence
Multimodal support (audio, images, video)
Enterprise cloud deployment options
99.5%+ uptime SLA

Pros

Significantly faster inference speeds
Lower cost per task (up to 70% reduction)
Frontier-quality reasoning
Drop-in replacement for traditional LLMs
Fine-grained output control
Enterprise-grade reliability and privacy

Cons

Early access/limited availability for some models
Newer technology with less production history than established LLMs
Requires understanding of diffusion-based generation paradigm

Use this if

You need ultra-low latency LLM inference, want to reduce AI infrastructure costs, are building real-time voice agents or code editors, or require high-throughput multi-step AI workflows.

Skip this if

You require maximum model size or reasoning capability regardless of latency, need models with extensive production history, or prefer traditional autoregressive generation paradigms.

Best for

Real-time voice agentsCode completion and editingComplex reasoning tasksHigh-volume inference workloadsCost-sensitive applicationsMulti-step AI workflows

Alternatives

OpenAI GPT-4Anthropic ClaudeGoogle GeminiMeta LlamaMistral AI

Compare Inception alternatives

More AI

Compare all

You.com Web Search APIs

AI-ready web search APIs delivering real-time results, content extraction, and grounded answers for agents and LLMs.

Amp

Coding agent and development environment that runs autonomous agents in cloud orbs or locally via CLI.

E2B

Isolated cloud machines for AI agents to run code, browse, use tools, and execute tasks at scale.

Browserbase

Cloud infrastructure for AI agents to browse and interact with the web like humans.